ATWZ (Agent Team Work Zone), a filesystem-based operations layer built around Claude Code's native Agent Teams that addresses problems and substantially mitigate the agentic"technical debt" described above.
Abstract
Large Language Model (LLM) agents have significantly improved coding and programming workflows. Claude Code, in particular, is one of the most powerful LLM coding agents and is capable of conducting complex coding tasks. However, several drawbacks can undermine long-term agentic workflows. (1) Irrecoverable agent teams: The Agent Teams feature is powerful, but the working state accumulated by each teammate is lost and cannot be resumed once the process stops, for example, when a terminal is closed. (2) Compaction erodes working detail: Compaction condenses the conversation into a summary, causing an agent's working details to become vague. (3) Agentic"technical debt": Over time, a user's decisions and the agents'operations become trapped in compacted old chats, making the project increasingly difficult to maintain and review. (4) Heavy prompt writing: Assigning or handing off tasks requires users to repeatedly write long prompts to achieve the expected agentic performance. We propose ATWZ (Agent Team Work Zone), a filesystem-based operations layer built around Claude Code's native Agent Teams that addresses these problems. Its central design principle is to treat each agent and teammate as a human employee and preserve their important working state in files stored in a dedicated directory called a"workstation,"together with the skills, hooks, and scripts that use and maintain these files. With ATWZ, an agent team can periodically back up its working state, allowing an agent's knowledge to be recovered after compaction. After a process ends, the team can be restored with a single command. These features also substantially mitigate the agentic"technical debt"described above. Moreover, within ATWZ, agent"employees"can send documents to one another, greatly reducing the effort required to write prompts.
AgentRadio is presented, an asynchronous message-passing layer that equips coding-agent harnesses with three primitives: threads, messages, and waiting for mentions that shows the gain growing with task difficulty, consistent with mid-course correction as the underlying mechanism.
Xinxing Ren, Qianbo Zang, Ziyan Wang et al.· 0 citations
Concurrent multi-agent coding promises division of labor across modules, robustness through redundancy, and parallel exploration at the natural granularity of multi-file projects. Realtime collaborative editing protocols solve this coordination problem for human teams via Conflict-free Replicated Data Types (CRDTs), but the LLMs underneath generate one token at a time and existing multi-agent coding systems inherit this serial limit: they either sequence agents through phase handoffs or pool independent samples without coordination, and a single agent abandons up to half of hard tasks with a one-file stub-and-exit. AgentRoom is a realtime collaborative editing protocol for concurrent coding agents. Its runtime layer exposes file-level claim, status, and broadcast as MCP tools on a CRDT-merged shared filesystem. Five frontier coding-CLI models ran four backend coding tasks, with cross-language checks in Python DevBench and Rust+axum. For CLI-stable models, AgentRoom with 2 agents abandons fewer tasks than Solo and has less run-to-run variation. At matched-compute, one positive mean LLM-judge contrast puts AgentRoom over parallel-merge. The other contrast, a bundle probe, puts full AgentRoom above each partial case: an ordering rather than a percentage split. Coordination, not parallelism or CRDT-merge, bears the load.
A ScholForge - hierarchical multi-agent LLM system that tackles end-to-end software engineering research end-to-end, with a natural-language research goal, and produces working code, experimental results, and a compiled manuscript without human intervention.
A. Stepin, Boris Tolstokulakov, V. Kulikov et al.· SIGSOFT FSE Companion· 0 citations
This paper synthesizes 27 benchmark, taxonomy, and audit papers (2023-2026), spanning 19 distinct benchmarks, into a cross-cutting taxonomy of agent limitations, the first synthesis that integrates evidence across tool use, planning, long-horizon reasoning, multi-agent coordination, safety, and measurement validity into a single, unified taxonomy of LLM agent limitations.
Wael S. Albayaydh, Rui Zhao, Ivan Flechais· 1 citation
Agent Skills package procedural instructions and checks for use by general-purpose agents, but loading a skill does not guarantee that every requirement remains active throughout a long tool-using trajectory. We study this problem in a production-derived, white-box code-audit workflow. Holding the task and 24 artifact checks fixed, we vary the surrounding context and classify where failures first become visible: lost requirements, editing drift, failed checking, or non-agent evaluator/runtime failures. Codex with gpt-5.4-mini passes 8/10 runs in a 10,991-character clean context but only 3/10 in both a 299,140-character relevant context and an equal-length irrelevant context. This 50-percentage-point difference is large but remains trend-level under two-sided Fisher tests (p = 0.0698). Requirement coverage nevertheless stays above 92% in both long conditions, showing that a few omissions can invalidate an otherwise complete artifact. A second task passes all clean and long runs, so the evidence does not support a universal context-length threshold. A detailed external checklist passes 10/10 runs, compared with 5/10 for a generic self-check (p = 0.0325). Coding-agent scaffolds may help by selecting a smaller working set, but they do not eliminate failures. We do not introduce context rot or a new general monitoring method; we provide a bounded failure classification and empirical case study for white-box code auditing.
Maintaining open-source repositories demands continuous attention to issue triage, code review, build monitoring, and community support—tasks that overwhelm individual maintainers when existing automation tools operate in isolation without shared context. Static analyzers check code quality, CI/CD systems run automated tests, and simple bots handle basic labeling, yet none of these tools share information with each other, leaving maintainers as the sole integration point between disconnected systems. We present Repogent, a multi-agent system where four specialized AI agents—Issue Manager, Pull Request Reviewer, CI/CD Maintainer, and Community Assistant—collaborate through event-driven coordination and persistent cross-task memory. Our GitHub webhook-driven architecture processes events through a priority queue, routes them to specialized agents via intelligent orchestration, and stores decisions in a persistent context layer that enables cross-task reasoning. Events are assigned CRITICAL, HIGH, or NORMAL priority, while a semantic memory module based on sentence-transformer embeddings supports code-level search, and all agents share a common LLM service using Qwen 3 32B accessed through the Groq API. Evaluation on a dataset of 150 issues, 100 pull requests, 80 CI/CD workflow executions, and 120 community queries collected from five active open-source Python repositories demonstrates that Repogent achieves 93.3% accuracy in issue classification, 69.0% combined review coverage for pull requests, 86.7% accuracy in CI/CD failure categorization with a macro F1-score of 0.884, and an average quality score of 3.96 out of 5 for community responses. These results establish a foundation for sustainable open-source maintenance through context-aware, multi-agent workflow automation. Furthermore, the proposed framework demonstrates that multi-agent orchestration with persistent shared memory can serve as an effective and scalable solution for automating repository maintenance activities, improving coordination across heterogeneous development workflows, reducing maintainer overhead, and enabling more consistent decision-making across repository events.
Venkata Satya Santhi Somisetti, Vijaya Bhaskar Santhuluri, Sai Teja Pathivada et al.· International Journal of Lat...· 0 citations
